Self-instruction can be an important mediator of generalized cognitive functioning. These studies evaluate the effects of self-instructions on one aspect of children's cognitive functioning, problem-solving. The target skill is the in-common concept. Children are asked to sort stimulus cards into 2 groups: those that have what a sample pair of complex stimuli have in common, and those that do not. The putative task analysis of that generalized skill will be taught: naming common features; sorting according to instructed features, and then according to instructed common features; etc. In certain children, this task analysis is not sufficient to produce generalized accurate sorting, but evoking a self-instruction by asking, "What are you looking for?" immediately makes that analysis sufficient. Our specific aims now are, first, to better understand the conditions under which this self- instructional effect operates, and to extend its applicability to subjects with a variety of developmental delays: (1) to ask if the self- instructional component is as necessary as it now appears, in these children; (2) to ask what can make a correct naming into a mediating naming; (3) to determine whether what the children say in this task mediates their solution or merely describes it; (4) to assess to extent of generalization of the self-instructional effect to novel tasks; (5) to assess the availability of self-instructional facilitation in less capable subjects; and (6) when less capable subjects do not benefit from verbal self-instruction, to develop nonverbal means of accomplishing a similar self-instruction can have in facilitating problem solving, and also suggest ways to improve such skills in young children and people with mental retardation.